Traction Synchronous Motors with Rotor Field Winding: A Literature Review

Abstract

Synchronous motors with a field winding in the rotor, known as wound-rotor synchronous motors (WRSMs) or electrically excited synchronous motors (EESMs), are claimed to be a good alternative to induction motors and even permanent-magnet synchronous motors (PMSMs) in electric traction applications. WRSMs do not require expensive rare-earth magnets and potentially have high power and torque density, and lower inverter power and cost, especially in applications demanding a wide constant-power speed range. Designing WRSMs for electric traction imposes some challenges and requires careful analysis. This paper provides an overview of commercial WRSMs for ground electric transport over the past 40 years, a comparison of WRSMs with other types of electric motors suitable for electric traction, and an overview of optimization methods and brushless excitation technologies for such machines. The goals of this paper are to present and discuss design approaches for traction WRSMs, to benchmark WRSMs against other motor types used in ground electric transport, and to highlight the most promising WRSM topologies and design techniques.

1. Introduction

According to a report by the International Energy Agency (IEA) [1], global sales of electric vehicles will exceed 17 million in 2024, accounting for more than 20% of the market. It is driven not only by the light vehicle segment but also by heavy-duty traction. Thus, electric truck sales inflated by nearly 80% in 2024 and are approaching 2% of total truck sales [1]. The global electric vehicle fleet is set to grow twelvefold by 2035 [1].

The majority of hybrid and battery light electric vehicles (EVs) historically employed permanent magnet synchronous motors with rare-earth magnets. Heavy-duty traction has been reliant on induction motors (IMs). Also, IMs can be found in a minor share of light EVs. A relatively new trend in light EVs is the adoption of electrically excited (wound-field rotor) synchronous motors. As shown in this paper, WRSMs are used in traction drives of various capacities from tens of kW to several MW. Wound-rotor synchronous motors (WRSMs) do not contain expensive rare-earth magnets and are not dependent on unstable prices and supply chains, making them a promising choice nowadays. Additionally, electrical excitation provides another degree of control freedom, enabling easier weakening of the magnetic field required for EVs to operate at high speeds. Particularly, it has a systemwide effect because it influences not only the efficiency of the machine itself, but also inverter power and cost. Heavy-duty vehicles can potentially benefit from it even more than light-duty vehicles, since one of the requirements in many applications is to provide a wide constant-power speed range (CPSR). It is caused by high power traction demand at a very low speed, combined with a light load at a relatively high speed required for vehicle transportation.

The advantages of WRSMs are closely tied to their drawback, which is the need to feed the rotating excitation winding. It requires an additional unit in the machine containing slip rings and brushes, or a brushless exciter.

The review includes a brief historical perspective of WRSM applications in ground electric transport since the 1980s and covers railway vehicles and light road vehicles with WRSMs. Next, a review of comparative studies on various electric motor types suitable for electric traction is provided, based on the available scientific literature. Design optimization procedures and results reported in the scientific literature are also presented. Brushless excitation techniques are reviewed in the corresponding section as an important component that can enhance the wider applicability of WRSMs in electric traction. Finally, a summary and conclusions are provided.

The research methodology of this study consists of searching and comparatively analyzing the publicly available scientific literature and disclosed manufacturer information such as datasheets and patents, on selected aspects of the topic under consideration. The search scope of this study is narrowed to the wound-rotor synchronous motors with field windings placed in the rotor (WRSMs, also referred to as electrically excited or wound-field synchronous motors—EESMs and WFSMs) used exclusively in electric traction applications, searched as traction, electric vehicle, hybrid vehicle, electric mobility, and automotive motors. Non-traction motors are not considered due to their different design objectives and duty cycles, and typically lower electromagnetic loadings.

Section 2 presents examples and applications of WRSMs in production traction systems from the 1980s to the present. The earliest traction WRSM references trace to Alstom high-speed trains developed in the 1980s, which are included in Section 2 mainly for historical context. The main time coverage embraces 2012–present time, as WRSMs for EV traction studies emerged in 2012–2015 in parallel with WRSM expansion in the EV industry, also reflected in Section 2, with most publications appearing from 2021 onward.

Section 3 presents a literature review comparing WRSMs with other motor types in traction applications. Comparative studies are included in this section only if WRSM benchmarking was performed for traction applications against the most competitive solutions, such as permanent-magnet synchronous motors, permanent-magnet-assisted synchronous reluctance motors, or induction motors.

Section 4 reviews the methods for the optimal design of traction WRSMs. In this section, priority is given to the studies employing optimization-based design techniques as the most trustworthy and proven to achieve optimal performance. At the same time, studies devoted to narrow theoretical aspects not explicitly related to the design of traction WRSMs, modeling techniques, control-focused studies, and studies devoted to specific design features which had not been proved by experimental verification are not reviewed.

In addition, since, as it is shown in Section 2, the field winding on the WRSM rotor in traction applications is typically powered through a sliding contact, which creates certain difficulties and limits the range of applications of such electric machines, in Section 5, we consider various brushless excitation methods proposed specifically for traction WRSMs and draw conclusions regarding their prospects.

2. Commercial Traction Wound-Rotor Synchronous Motors for Ground Vehicles (1980s–Present Time)

This section discusses practical examples of WRSM implementation in production vehicles. As the flowchart in Figure 1 shows, all the examples found in the literature can be divided into two broad categories: railway rolling stock and automotive applications.

2.1. Railroad Traction

One of the first known facts of WRSM commercial application for ground vehicles refers to 1988, when Alstom began production of a high-speed train originally called ‘Train à Grande Vitesse Atlantique’ (TVG-A) [2]. The production lasted until 1992 by Alstom in France and was continued by CAF for the Class 100 localized version in Spain [3]. The train was driven by eight forced-air-cooled WRSMs, each with a continuous power of 1100 kW. The maximum commercial speed of the TVG-A train was 300 km/h. The corresponding maximum WRSM speed is estimated at 3982 RPM based on available data [4] consisting of the wheel diameter of 875 mm and a gear ratio of 1:2.1894. The version of the TVG-A train that set the world record of 515.3 km/h in 1990 had a WRSM with a maximum speed of around 4600 RPM due to modifications to the wheels and gearboxes. The weight of one motor was 1450 kg. A salient-pole rotor had six poles with slip ring contacts to feed the field winding.

A WRSM was selected for the TGV-A train instead of an IM based on the techno-economic assessment described in [5]. Electric drive with an IM was rejected because of the higher complexity, cost, volume, and weight of the thyristor inverter [5], which required forced commutation while supplying an IM. The WRSM of TGV-A was fed by a six-pulse thyristor bridge [4,6]. However, according to [6], TGV-A utilized thyristors in a simple bridge configuration able to work as an inverter where thyristor commutation was achieved by a ‘load commutation’ caused by the WRSM’s back EMF as a leading power factor.

Later, TGV Réseau and the first series of TGV-Duplex were developed based on TGV-A [7] and were manufactured during 1992–1996. An export version of TGV Réseau was manufactured in South Korea under the name of KTX-I from 1997 to 2003. An enhanced version of the WRSM with 1130 kW of continuous power was employed [8].

Also, Alstom began production of Sybic BB class 26000/26500 locomotives in 1988 [9]. The production lasted until 1998. The maximum speed was 200 km/h. The locomotive had two WRSMs with a continuous power of 2800 kW and the maximum speed of 1930 RPM [9]. The motor had eight poles and was combined with the gearbox with a 1:2.1894 ratio [9]. It is reported that the motor had dual three-phase winding shifted by 30 degrees [6].

Consequent Alstom trains employed cheaper IMs, which became advantageous because of the development of a new generation of inverters at that time. Particularly, it allowed for the elimination of brushes and slip rings in the traction drives. According to open sources, a significant number of the aforementioned trainsets and locomotives equipped with WRSMs are currently in commercial operation, which confirms their reliability and durability. Such railway vehicles are perfect examples of high-power traction. Forty years of their successful application paved the way for other similar scenarios of heavy-duty traction.

The return of WRSMs to electric transport occurred in the second decade of the 21st century, when they demonstrated themselves as a more sustainable and cost-effective alternative to permanent magnet synchronous motors (PMSMs).

2.2. Road Vehicles

Renault Group introduced the concept of a battery-powered electric car named ZOE in 2010 [10]. The delivery began in 2012. A prominent feature of the Renault ZOE is a WRSM with forced-air cooling and brushed excitation supply [10]. Sometimes, this type is called ‘synchronous motor with coiled rotor’ in Renault specifications [10]. The same principle is used in the following models of Renault cars, such as Twingo, Kango, Megane, Scenic, and Fluence [11,12]. Several car manufacturers, namely BMW, Nissan, Rolls-Royce, Mini, and Smart, also employ WRSMs in their vehicles, according to [11], often calling them externally or electrically excited synchronous motors (EESMs). The maximum power already reaches 360 kW, while the maximum torque is 710 N∙m and the maximum speed is 17,000 RPM.

The BMW Group began selling the BMW iX3 model in 2021. The powertrain is based on WRSMs [12]. It refers to a current-excited synchronous electric motor in BMW specification [13]. The BMW Group follows the strategy of avoiding rare-earth magnets in electric motors [14]. Thus, the BMW Group website [15] states that the new 6th generation of the powertrains will adopt WRSMs as well. BMW owns numerous patents related to WRSM manufacturing. For instance, a manufacturing method of a salient-pole rotor is described in [16]. The proposed method enhances structural integrity and allows for higher speeds to be reached. Similarly, patent [17] is also devoted to rotor strength enhancement. Patent [18] suggests a shaft composed of two parts: a shaft body and a current transmission module, which helps to withstand higher speed. A winding retention device for a wound rotor based on a star disk shape is introduced in [19].

Nissan Group offered a fully electric vehicle named Nissan Ariya with WRSM in 2022 [20,21]. Rolls-Royce introduced the first electric Rolls-Royce Spectre in 2023 [22]. It features two WRSMs with a total power of 550 kW and a maximum output torque of 1075 N·m. The motor is called a ‘Separately Excited Synchronous Motor (SESM)’ in Rolls-Royce specifications. MINI, owned by the BMW Group, also adheres to the policy of avoiding rare-earth magnets in electric vehicles [23]. The MINI Countryman E and MINI Countryman SE ALL4 with WRSMs were offered in 2024. Electrical versions of Smart employ WRSMs from Renault [24] with 41 kW of continuous power and 60 kW of maximum power.

The Nissan/Renault Alliance patent portfolio, in particular, consists of cooling solutions, such as [25], which introduce oil-spray cooling of the rotor end windings and mechanical ‘collars’ at each end of the rotor. Each collar has radial discharge ducts that drain sprayed oil from the hollow end region and direct it toward the stator coil heads, preventing oil ingress into the air gap and, therefore, limiting drag losses. Various mechanical solutions are presented for the rotor winding retention and for enhancing the rotor mechanical strength [26,27,28,29,30,31,32]. A segmented cage encasing the wound rotor stabilizes the coil sides and end-turns, reducing the number of parts and simplifying assembly, as presented in [26]. Complementing this, a wound-type rotor structure introduces guide heads and stack-end features that improve the routing and balancing of the conductors, as shown in [27], while slot-closure wedges and pole shoe interfaces for locking the field winding under centrifugal load are presented in [28]. Three related guiding devices are presented: a shaft-mounted winding guide for salient poles, which improves path control and protects insulation during needle-winding in [29]; a coil-head guide which defines the rotor winding overhang geometry in [30]; and a general guiding device for rotor winding designed to enable automated placement in [31]. A wire-guiding head that shrink-fits to the shaft, providing retention at high speed, is presented in [32].

ZF Friedrichshafen AG announced the upcoming adoption of traction WRSMs with brushless excitation. The manufacturer names it the ‘in-rotor inductive-excited synchronous e-motor (I2SM)’. An inductive brushless exciter is located in the hollow shaft [33]. It is stated that the power ratings in the production series of motors are expected to be from 75 to 550 kW to serve various transportation scenarios [33,34]. However, more detailed motor data are not available. Some information can be found in ZF patents. Thus, in [35], a rotary transformer and rectifier are placed inside the shaft and with an axial press-fit conductive interface between the transformer secondary and the rectifier to deliver DC to the field winding; patent [36] presents an improvement of the arrangement of the rotor modules inside the shaft, detailing compact axial stacking and provisions for cooling and earthing to minimize length and ease of assembly while maintaining balance at high speed; and patent [37] integrates a contactless exciter with rotor earthing, a rotor-position sensor, and oil routing through the shaft, so the exciter, rectifier, and sensor can coexist without parasitic effects.

Valeo and Mahle announced a plan to establish joint mass production of traction WRSMs with power from 220 to 350 kW and brushless excitation [38]. Valeo owns various patents related to WRSMs. Thus, energy-saving rotor field control is claimed in [39]. Patents [40,41] describe a WRSM rotor cooling system comprising a central shaft channel and channels located in each pole, which can be used in addition to oil-spray cooling. Rotor stacking and field winding assembly technologies are provided in [42].

Table 1. Application examples of permanent magnet-free WRSMs in ground-based electric vehicles (1980s–present time).

Model Name	Maximum Power, kW	Maximum Torque, N∙m	Maximum Speed, RPM	Ref.
Railway
TGV Atlantique	1300	NA	3982	[2,4,5]
KTX-I	1130	NA	4000	[8]
Sybic BB 26000/26500	2800	NA	1930	[9]
Automotive
Renault ZOE R135	100	245	11,300	[10,12,43]
Renault Twingo Z.E.	60	160	11,450	[11,43]
Renault ZOE R110	80	225	10,900	[11,43]
Renault Kango Z.E.	90	245	11,450	[11,43]
Renault Megane EV40	96	250	11,155	[11,43]
Renault Megane EV60, Scenic EV87	160	300	11,688	[11,43]
Renault Scenic EV60	125	280	8900	[11,43]
Renault Fluence Z.E.	70	226	11,000	[11,43]
Nissan Ariya FWD Empower+	178	300	13,520	[20,43]
Nissan Ariya Platinum+ e-4ORCE	290	599	13,520	[21,43]
Rolls-Royce Spectre	Forward: 190 Rear: 360	Forward: 365 Rear: 710	NA	[22]
MINI Countryman E	150	250	15,000	[23,43]
MINI Countryman SE ALL4	230	494	15,000	[23,43]
Smart EQ Forfour, Fortwo, Cabrio	60	160	11,450	[24,43]
BMW iX3	210	400	17,000	[12,43]
BMW X1 iX1 xDrive30	Forward: 140 Rear: 140	Forward: 247 Rear: 247	NA	[44]
BMW iX2 iX2 eDrive20	150	250	NA	[45]
BMW i7 M70 xDrive, iX M60	Forward: 190 Rear: 360	Forward: 365 Rear: 650	NA	[46,47]
BMW i7 xDrive60	Forward: 190 Rear: 230	Forward: 365 Rear: 380	NA	[48]
BMW iX xDrive40, xDrive50	Forward: 190 Rear: 200	Forward: 290 Rear: 340	NA	[49]
BMW i4, i5 eDrive40: Sedan, Touring	250	430	17,000	[43,50,51,52]

presents the main characteristics of the motors employed by the aforementioned manufacturers. In Table 1, the symbol “NA” in some cells indicates that data are missing. This means that the corresponding sources do not contain information on this value. Notably, the maximum torque cannot be calculated based on the maximum speed and maximum power, as maximum torque typically corresponds to low speeds beyond the CSPR. The plots in Figure 2 illustrate data for the WRSM traction drives listed in Table 1. Based on the data in Table 1 and Figure 2, it can be concluded that commercially produced WRMSs are used in light passenger vehicle drives with CPSRs in the range of approximately 2:1 to 3:1. Also, from an analysis of the sources listed in Table 1, it can be concluded that all commercially produced traction WRMSs use brush contact to power the rotor field winding.

3. Comparative Analysis of EV Traction Motors in the Scientific Literature

Multiple studies have been devoted to the benchmarking of various traction motors, including WRSMs. This section describes characteristics and metrics of WRSMs in comparison with other viable motors for electric traction, mainly PMSMs, Ims, and reluctance-based motors.

Benchmarking of various types of traction motors is provided in reviews [53,54,55,56] in a generalized form. The majority of analyzed studies compares WRSMs with PMSMs [56,57,58,59,60,61,62,63,64,65], sometimes including reluctance motors in the analysis. A couple of studies discuss only rare-earth free motors, including WRSMs [66,67,68].

3.1. General WRSM Evaluation Compared to Other Motor Types

Article [53] compares spoke-type ferrite interior PMSMs (IPMSMs), switched reluctance motors, IMs, different SynRM designs, and WRSMs. It is stated that SynRMs are highly efficient motors with low torque density, which makes their application in electric traction challenging. Slipping electric contact in the excitation circuit is the main disadvantage of WRSMs. Also, additional transistors are required to adjust the rotor current. The torque density is 10 N∙m/kg for both IM and WRSM; however, ferrite-based IPMSMs demonstrate 11 N∙m/kg.

A comparative analysis of IPMSMs, WRSMs, SynRMs, and IMs is presented in [54]. SynRMs are described as having low torque density and a limited constant-power speed range (CPSR). WRSMs do not expose such issues; however, the torque density does not reach the level of IPMSMs. Their main challenges are providing the supply of the rotor field winding and implementing the heat removal from the rotor.

Power density ranges for IMs, permanent-magnet-assisted SynRMs (PMaSynRMs), IPMSMs, and WRSMs are analyzed in [55] based on commercial models’ data. IMs typically have lower efficiency, especially at low speeds. However, their high torque at low speed and low maintenance requirements make them a good choice for heavy-duty and off-highway/off-road traction, as well as for frequent start–stop cycles. WRSMs, being rare-earth magnet-free, have a relatively low cost, a wide CPSR, and high torque at low speed.

IPMSMs, WRSMs, SynRMs, and IMs are included in the comparison in [56]. It is stated that WRSMs represent an overall balanced solution taking into account its reliability, efficiency, torque density, and cost.

3.2. Light Electric Vehicles

Currently, most EVs belong to the light vehicle class. Traction motors of this category rarely exceed 160–200 kW of maximum power, and performance is often tested for standard conditions such as WLTP or Artemis driving cycles. The typical constant-power speed range usually does not exceed 4:1–3:1.

A WRSM and an IPMSM for a light EV are analyzed in [57]. The maximum power is 225 kW and the maximum speed is 16,000 RPM. It is mentioned that WRSM is more sustainable, cheaper in production, and has a lower weight by 16%. However, average power losses for the WLTP cycle are 1.0 kWh/100 km for the IPMSM and 1.1 kWh/100 km for the WRSM.

A WRSM and an IPMSM for an A-class battery electric vehicle were designed in [58]. The rated voltage is 48 V, the rated power is 30 kW, and the peak power is 60 kW. The base speed is 4696 RPM and the maximum speed is 15,900 RPM. It is reported that the cost of active materials is twice as high for the IPMSM. The maximum WRSM efficiency is only 93% versus 96% for the IPMSM, but the losses at high speeds are significantly higher for the IPMSM. A valuable part of [58] is the comparison for common driving cycles, WLTP3, Artemis Urban, and Artemis Motorway 130. For the city-dominating cycle, Artemis Urban with the average vehicle speed equal to 17.7 km/h, average cycle power losses are smaller for the IPMSM. However, WRSM demonstrates significantly lower average cycle power losses for the WLTP3 and especially for the Artemis Motorway 130 cycle, which assumes an average speed of 96.9 km/h and a maximum speed of 131 km/h.

Article [59] provides a theoretical benchmarking of an IPMSM and a WRSM with the same outer stator stack diameter of 220 mm and axial length of 128 mm. The motor is supposed to be used in a mid-size light EV with a curb weight of 2200 kg. Maximum torques achieved for the IPMSM and the WRSM are 380 N∙m and 320 N∙m, respectively. The maximum speed is 14,000 RPM, base speed is 5500 RPM for the IPMSM, and 5000 RPM for the WRSM. The torque density of the WRSM is lower by 16%, which is caused by extra losses in the rotor excitation winding. The average efficiency of WRSMs is lower by 8.3% for the WLTP cycle. The difference is just 3.3% for the Artemis Motorway 150 cycle, assuming a maximum speed up to 150.4 km/h and an average speed of 99.6 km/h.

Article [60] is devoted to the detailed comparison of IPMSMs with rare-earth and ferrite magnets, a PMaSynRM with ferrite magnets, and a WRSM. All motors are designed for a maximum speed of 12,000 RPM and a maximum torque of around 400 N∙m. Equalizing the maximum torque across motors is achieved by the axial length variation, which led to different base speeds for each motor. The maximum power is different: it is 162 kW for the WRSM, 228–236 kW for the IPMSMs with different rare-earth magnet grades, 126 kW for the IPMSMs with ferrite magnets, 135 kW for the PMaSynRM with rare-earth magnets, and 102 kW for the PMaSynRM with ferrite magnets. Also, the resulting CPSR varies. According to [60], a WRSM has a narrower CPSR than rare-earth-enabled IPMSMs but exceeds all PMaSynRMs and ferrite IPMSMs. The WRSM demonstrate moderate WLTP driving cycle efficiency; however, it exceeds the rare-earth IPMSMs. All reluctance motors and ferrite IPMSMs demonstrate noticeably higher efficiency.

Thermal performance is also investigated in [60] for a single speed of 6000 RPM. Spray cooling is simulated; however, the field winding temperature can reach up to 143 °C, highlighting the need for more intensive spray cooling or additional shaft cooling. Also, structural integrity analysis is performed in [60]. A maximum speed of 14,400 RPM is set for stress analysis. WRSMs demonstrate the highest levels of stress and deformations and the lowest safety factor equal to 0.76 among all of the studied motors. The safety factor reaches 2.4 for IPMSMs and 1.26–1.35 for PMaSynRMs. The final integrative rating of the WRSM is the highest in [60]. It included electromagnetic, mechanical, cost, and sustainability ratings.

Study [61] compares simulation results for an IPMSM, an IM, and a WRSM with the same maximum torque of 430 N∙m, maximum power of 270 kW, and maximum speed of 15,000 RPM. It is reported that IM is 50% heavier than the IPMSM and the WRSM, which both have similar weights. Also, IM have a larger diameter of 254 mm versus 220 mm for its counterparts. Both magnet-free motors (IM and WRSM in [61]) need special and more intensive techniques for rotor cooling (for example, spiral shaft groove) to achieve the same performance as the IPMSM. A stator water jacket was supposed for all motors.

A model-based analysis is performed in [62] for a WRSM, a hybrid-excited WRSM (HeWRSM), a SynRM, and a PMaSynRM. All motors are designed for the same rated torque of 200 N∙m, rated speed 4000 RPM, and maximum speed 20,000 RPM. Also, the outer stator stack diameter is equal to 242 mm, and the axial length is 132 mm. The analysis shows that the torque-speed characteristics of the IPMSM, HeWRSM, and WRSM lie in the vicinity of each other above the rated speed. The SynRM curve lies lower, demonstrating a narrow CPSR. The PMaSynRM curve lies in between. The efficiencies of the SynRM and the PMaSynRM are high though. The estimated SynRM cost is the lowest, which is more than two times cheaper than the IPMSM. The WRSM cost is moderate, accounting for 1.32 times the cost of SynRM but still 1.75 times cheaper than the IPMSM.

A detailed analysis is provided in [63] for a WRSM and a PMaSynRM with ferrite magnets. The specification assumes the same maximum core dimensions: 220 mm of outer diameter, 140 mm of axial length, and the rated power of 150 kW at 6000 RPM. The maximum speed is 14,000 RPM. WRSM’s efficiency is lower across the operating range by 0.6–1.4% depending on the operating point. Both the WRSM and PMaSynRM have equal power factors at the maximum speed. The power factor of the WRSM at the rated speed and peak power is much higher and equal to 0.95, while the power factor of the PMaSynRM is only 0.71. The difference in power factors affects the phase current at the rated speed and peak torque (423 A for the IPMSM and 328 A for the WRSM).

3.3. Electric Trucks

The truck applications discussed in [64,65,66,67,68] impose specific requirements such as high starting and low-speed region torques [65] and wider CPSRs which can reach 1:10 in some cases [66].

An optimization-based comparison of an IM and a WRSM specifically for a 90-ton dump mining truck is carried out in [66]. The required rated power is 370 kW. The CPSR of this application is very large (1:10), spanning from 400 RPM to 4000 RPM, which is far beyond the 1:3–4 ratio typically found in road vehicles. As a result, the active weight is lower by 15% for the WRSM and the active axial length is smaller by 17.5% compared to the IM, assuming a fixed outer stator stack diameter equal to 668 mm, as it was dictated by the mining truck specification. One of the main outcomes of [66] is that the WRSM’s phase current is lower by 1.3–1.5 times across the operating speed range, which is beneficial for the inverter power and corresponding cost in comparison with the IM. The total losses are lower by 9–24% for the WRSM rather than for the IM. An important issue is addressed in [66] as well which is discussed further. It was observed that high-amplitude low-frequency temperature fluctuations occur in the inverter supplying the IM when the truck is stopped on a slope, utilizing the torque developed by the motor at zero speed. Such a holding mode is required for the operators’ convenience. The amplitude of these fluctuations is relatively high, which results in significant inverter oversizing in IM-based drives. Inverters, feeding WRSMs, are not susceptible to this issue due to a different holding torque generation principle. Thus, WRSMs can hold the vehicle on a slope by DC supplied to the stator winding, unlike IMs, where a low-frequency AC supply is required to develop the holding torque [66]. Taking into account frequent stopping on a slope for this application, the absence of temperature fluctuations in the inverters of WRSM-based drives significantly increases the powertrain’s reliability.

A synchronous homopolar motor (SHM), having excitation winding in the stator, is compared with a WRSM in [67] for the same specification and application requirements as in [66]. Both machines have the ability to adjust the excitation current. WRSM’s efficiency is superior by 0.5–0.9% depending on the working point; it demonstrates average duty cycle losses of 22.7 kW, which is noticeably lower than 27.4 kW for the SHM. Also, the phase current amplitude is lower for the WRSM, especially for field-weakening operation. The difference becomes twice as high at the maximum speed. Moreover, the WRSM has lower torque ripple. However, SHM construction assumes a simple and robust rotor without windings, which is easier to manufacture with such a large diameter, taking into account the maximum speed of 4000 RPM.

In [68], a SHM and an IPMSM are compared when used to drive the same mining truck. A WRSM is not mentioned in [68]; however, the considered motors are optimized with the same specification as in [66,67]. Thus, these IPMSM and WRSM can be compared considering their comparison with SHM. Thus, the IPMSM’s current amplitude reaches 631 A at the maximum speed, which is more than two times higher than for the WRSM at the same speed. IPMSM’s efficiency accounts for 86% at the maximum speed, while for the WRSM it is 95.6%, highlighting the drawback of IPMSMs in applications with a large CPSR. At low speed and rated torque, IPMSM’s efficiency is higher by 0.8%. The active weight of the IPMSM is lower by 11.6%; however, the estimated cost is 3.6 times higher because of the expensive rare-earth magnets.

A study of WRSM and IPMSM applicability for a 40-ton long-distance heavy truck is provided in [64]. Required rated and maximum speeds are 3000 RPM and 8000 RPM, respectively. The maximum power and torque are 215 kW and 581 N∙m. The outer stator stack diameter of 258 mm and the axial length of 200 mm are the same for both machines. The truck’s duty cycle consists of around 90% of high-speed operations. In such high-speed conditions, the WRSM’s efficiency is higher: 92.46% versus 87.61% for the IPMSM; however, for high loads at low speed, it is the opposite. Average driving cycle efficiency is better for the WRSM by 4.11%. Also, the total weight of the WRSM is higher, but the total material cost is lower. The power factor reaches 0.96 for the WRSM, being only equal to 0.45 for the IPMSM. Expectedly, it leads to large differences in electrical losses and in the current consumption for an inverter.

The same driving cycle as in [64] is considered in [65], for a slightly different specification: the maximum power and torque are 250 kW and 800 N∙m. The maximum speed is 10,000 RPM and the rated speed is 5041 RPM. Outer stator stack diameter is 300 mm and the axial length is 360 mm. Provided theoretical results confirm that the WRSM demonstrates better efficiency at high speeds and low loads than the IPMSM. Also, it is highlighted that the WRSM’s phase current amplitude is 3.67 times lower at the maximum speed than for the IPMSM.

3.4. Summary

Table 2. Main motor parameters in the benchmarking studies [57,58,59,60,61,62,63,64,65].

Application	Motor Ratings	Ref.
Light electric vehicles
D–E-class EV	Maximum power 225 kW	[57]
A-class battery EV	Rated power 30 kW, maximum power 60 kW, maximum torque 122 N∙m, CPSR 3.39:1	[58]
Mid-size light EV	Maximum power 160 kW, maximum torque 320 N∙m, CPSR 2.8:1	[59]
WLTP class 3 EV	Maximum power 162 kW, maximum torque 400 N∙m, CPSR 3.16:1	[60]
Light battery EV	Maximum power 270 kW, maximum torque 430 N∙m, CPSR 2.73:1	[61]
Light EV	Rated power 90 kW, rated torque 200 N∙m, CPSR 5:1	[62]
C-class EV	Rated power of 150 kW, rated torque 240 N∙m, CPSR 2.33:1	[63]
Electric trucks
90-ton hybrid electric off-highway truck	Rated power 370 kW, maximum torque 8833 N∙m, CPSR 10:1	[66,67,68]
40-ton long-distance heavy electric truck	Maximum power 215 kW, rated power 42 kW, maximum torque 581 N∙m, CPSR 2.67:1	[64]
40-ton long-distance heavy electric truck	Maximum power 250 kW, rated power 50 kW, maximum torque 800 N∙m, CPSR 1.6:1	[65]

represents the main motor parameters for the studies discussed in Section 3.2 and Section 3.3. Figure 3 summarizes the pros and cons of the most commonly used types of traction motors, as discussed in Section 3.

4. Design Optimization of Wound-Rotor Synchronous Motors

Multidisciplinary optimization plays a major role in electric machine design. This section collects data about techniques and objectives used for traction WRSM’s optimization.

A 370 kW WRSM (Figure 4) for a dump mining truck is optimized in [67]. The main motor parameters are given in Section 3. The optimization is performed for three points of the working cycle, which include the rated (base), maximum speed, and one intermediate point within the constant power region. The Nelder–Mead method is used for optimization. The objective function is built to minimize the average losses in the cycle, the phase current amplitude, and the torque ripple. The number of poles, the outer stator diameter, axial length, and rotor winding slot-fill factor are fixed. Seven geometrical parameters and the current angle are varied. The optimization results in reduced excitation and stator currents at the rated speed, which led to a significant decrease in average cycle losses, especially in the inverter power rating by 2.77 times.

The Solid Isotropic with Material Penalization method is used in [69] for topology optimization of a WRSM. The objectives are average torque maximization and rotor loss minimization for the rated speed of 4000 RPM, with the torque ripple constrained below 10%, and the rotor current density within the limits for oil-spray cooling and the mechanical stress safety factor. The stator dimensions are fixed: outer diameter of 254 mm and axial length of 106 mm. The topology optimization assumes that each finite geometrical volume is either empty or filled by the material. It is highlighted that electromagnetic topology optimization often delivers geometries that are difficult to manufacture or even completely non-existent, such as separated pieces of the pole shoe not connected with the main body, etc. Therefore, electromagnetic and structural optimization should be interconnected.

Article [70] offers a novel optimization method based on a combination of the maximum entropy sampling algorithm and the physics-informed Bayesian optimization algorithm. The algorithm is used for the electromagnetic design of a WRSM with an outer stator stack diameter of 240 mm, axial length of 110 mm, rated speed of 6000 RPM, and a maximum speed of 15,000 RPM. The objectives are to maximize the torque above 250 N∙m and power above 80 kW, maintaining a stator winding temperature below 150 °C and efficiency above 90%. Only stator geometrical parameters are adjusted, including the dimensions and shape of the slot conductor, which aims to maximize the slot-filling factor. After the optimization, a simplified thermal model is used to estimate the stator winding temperature. It is stated that the developed algorithm demonstrates computation efficiency 45% better than genetic algorithms common in the field of electric machines.

A WRSM in [71] has the following specifications: rated power of 160 kW, rated speed of 5300 RPM, rated torque of 280 N∙m, and maximum speed of 11,500 RPM. Optimization goals consist of total loss minimization and average torque maximization. The constraints are the rotor stress safety factor above two at 120% of the maximum speed, torque ripple less than 4%, and a thermal loading coefficient below 2.1 A²t/m³. The thermal loading coefficient is selected by assuming stator water jacket cooling. The efficiency is evaluated in the WLTP cycle for all design options. Stator cores made from silicon steels and soft magnetic composites (SMCs) are compared. According to [71], SMCs are considered cheaper and eco-friendly options, delivering similar performance to the silicon steel cores. However, detailed mechanical and thermal analyses of the SMC cores’ impact are not presented.

In [72], a six-pole WRSM with an outer stator stack diameter of 230 mm and axial length is designed in two versions: one with hairpin and one with pull-in winding. The proposed rated speed is approximately 6000 RPM and the maximum speed is 14,000 RPM. The objective function includes efficiency maximization in several load points between 7000 and 10,500 RPM, maintaining the torque. The global response search method is employed. Detailed AC loss analysis indicates that total losses of the motor with hairpin winding are lower by 6% than for the motor with pull-in winding. Hairpin winding allows for a 0.659 copper slot fill factor and pull-in winding allows for a 0.44 factor, which also affects the current density in both cases.

A multiple slot/pole combinations study is presented in [73]. The rated torque is 155 N∙m, the maximum torque is 350 N∙m, the rated speed is 4650 RPM, and the maximum speed is 15,000 RPM. The maximum stator stack diameter is set to 241 mm and axial length to 135 mm. The power density target is to exceed 5 kW/kg. Shaft cooling and oil spraying of the end windings and stator water jacket are required to achieve that. Ten geometrical parameters are varied using a genetic algorithm merged with an analytical model and a surrogate finite-element model (FEM) in order to process large generations of 1000 individuals over 100 iterations. WLTP efficiency and cost metrics for the optimized motors with several slot/pole combinations are obtained as a result.

Target characteristics in [74] are a peak power of 55 kW, rated power of 30 kW, rated speed of 4000 RPM, and maximum speed of 12,000 RPM for a WRSM with 48 slots and 8 poles. The first design step in [74] includes the search for mechanically stable ranges of the main rotor parameters using the central composite design of the experiment’s technique. Electromagnetic optimization employs a parallel stochastic differential evolution algorithm for the tuning of 12 geometric parameters and stator and rotor current densities. The aim of the optimization is to maximize the torque density and a ‘goodness’ parameter, which is defined as the average electromagnetic torque divided by the square root of total losses. The constraints are torque ripple, rotor electrical losses, stator total losses, average torque, current densities in stator, and rotor windings. End windings oil spraying is used in the prototype, which allows for the achievement of the power density of 1.95 kW/kg and the maximum power of 79.6 kW, confirming the effectiveness of this type of cooling.

The Toyota Prius IPMSM is selected as a benchmark in [75]. A WRSM is optimized to achieve 200 N∙m and 2800 RPM, considered as the rated point. Initial sizing is performed using an analytical model combined with finite element analysis (FEA), keeping a constant outer stator diameter, which leads, in turn, to the increased axial length. Then, parametric and topology optimization are used to maximize average torque and minimize torque ripple. The genetic algorithm manipulates eight stator and rotor geometrical parameters, resulting in 18.1% of extra torque. The result is achieved in 161 generations, having 90 individuals in one generation. Topology optimization involves only the rotor’s pole shape and has additional constraints such as maximum rotor stress and deformations. The torque increase accounts for 22.1% without mechanical integrity considerations and only 2% with imposed rotor strength requirements.

A stator from the Chevy Volt motor with hairpin winding was considered in [76] as a basis for WRSM optimization. The stator has 72 slots, 12-pole winding, a 260.86 mm of outer stack diameter, and 94.18 mm of axial length. The targeting rated and maximum power are 55 kW and 190 kW, respectively. The rated speed is 4000 RPM and the maximum speed is assumed to be 12,000 RPM. The objective function assumed maximization of the power density and driving cycle weighted efficiency, which consists of five points, including rated and maximum speed. The constraints are the torque, maximum voltage, stator and rotor current densities, maximum excitation power, and torque ripple. Since the stator parameters are fixed, only five rotor geometry parameters and the air gap are adjusted. The optimization algorithm consists of sensitivity analysis using Advanced Latin hypercube sampling and the Metamodel of Optimal Prognosis approach as preliminary steps, followed by Non-Linear Programming by Quadratic Lagrangian and an evolutionary algorithm. The best candidates are FEM-simulated in mechanical and electromagnetic domains to ensure the feasibility of the solution. The prototype is built with oil-spray cooling, square wire for field winding, and a segmented rotor. Weighted driving cycle efficiency reaches 94.45% and power density achieves a value of 37.8 kW/L. Two versions of the excitation supply are explored: a traditional brushed system and a contactless capacitive coupler. Details of the coupler design are described in [77] and in Section 5. Overall, technical report [77] provides a comprehensive review of a long-term WRSM development project with a contactless capacitive coupler, expanding publications [69,74,76] associated with that project.

The BMW i3 IPMSM with the rated torque of 150 N∙m at 4800 RPM and the maximum speed of 12,400 RPM is used as a reference in [78]. The electromagnetic optimization of a WRSM is based on a non-dominated sorting genetic algorithm (GA) combined with an analytical model and FE model. The objective is the efficiency maximization and torque ripple minimization at a single load point (9000 RPM, 40 N∙m) by varying geometrical parameters. The prototype of the WRSM is longer than the reference motor (180 mm vs. 132 mm) but has a slightly smaller outer stator stack diameter (230 mm vs. 242 mm), having the same rotor diameter of 179 mm. The number of slots is 48 and the number of poles is 4. As a result, better efficiency of WRSM is achieved for almost all working points than the reference IPMSM. The need for a pronounced saliency ratio and resulting reluctance torque is highlighted. Field-weakening techniques are also discussed in [78].

A multidomain optimization technique is revealed in [79]. It includes solving electromagnetic, mechanical, and thermal problems. The aim is to maximize the torque and efficiency in a single working point at medium speed and low load. The constraints are minimum torque (250 N∙m), minimum power (80 kW), minimum efficiency (90%), maximum temperature of windings (150 °C), and von Mises stress (maximum 200 MPa) caused by centrifugal force at the maximum speed of 15,000 RPM. The number of slots (48) and poles (8) is fixed, as well as the outer stator stack diameter of 212 mm and axial length of 200 mm. A combination of stator water jacket and oil-spray cooling systems is assumed for the analysis. Nondominated sorting GA is used to optimize eight rotor geometrical parameters. The number of individuals in one generation is 100, which are processed in one pass using a high-performance computing cluster. It is emphasized that many solutions lie beyond the structural or thermal limitations and pass only electromagnetic design criteria.

A WRSM with 54 slots, 6 poles, an outer stator stack diameter of 232 mm, axial length of 134 mm, 236 kW of maximum power, rated speed of 6000 RPM, and theoretical maximum speed of 18,000 RPM is optimized in [80]. The optimization is performed using FEA and GA for eight rotor geometric parameters and field current density. The goal is to increase weighted WLTP efficiency. The resulting design is compared with an IPMSM with the same dimensional specification.

Article [81] deals with GA utilization for torque maximization and torque ripple minimization for the rated speed of 5300 RPM condition. A WRSM has 54 slots, 6 poles, an outer stator stack diameter of 210 mm and an axial length of 100 mm. The expected maximum torque is 150 N∙m. Five variable parameters were selected after the sensitivity analysis. The final design demonstrates a torque of 175 N∙m and a torque ripple of 4.6%. It is noted that wider geometrical boundaries for the rotor pole shape lead to better characteristics during optimization.

An automotive WRSM with 48 slots and 8 poles is optimized in [82]. The design space is reduced to just two geometrical parameters after the sensitivity analysis. The optimization aims to reduce magnetic losses and acoustic noise caused by magneto-motive force harmonics. A 2D electromagnetic problem is solved with a 3D structural problem to define noise and vibration levels. One of the constraints was the maximum reduction in the average torque of less than 7%. The optimization is performed for three points of the Artemis driving cycle (3000, 5000, and 12,000 RPM). A lower level of magnetic losses and noise is achieved for certain speed regions as a result.

A hybrid-excited WRSM is studied in [83]. The rated power is 5 kW, the rated torque is 31.83 N∙m at 1500 RPM, the maximum speed is 3000 RPM, the number of slots is 72, the number of poles is 16, and the outer stator stack diameter is 236 mm. Rare-earth magnets are located along the d-axis. Sampling and sensitivity analysis are performed using the optimal Latin Hypercube method. Then, a surrogate model is built using the Kriging method. The optimization is performed by micro-GA. The objective is the minimization of the magnet dimensions, maintaining the torque and reducing the torque ripple. Four geometrical parameters related to the magnet size and surrounding area are optimized, which resulted in a 50% reduction in rotor electrical losses, 1.5 times lower field current density and twice lower torque ripple than for the non-optimized version.

Toyota Prius 2010 IPMSM is a reference motor in [84]. The WRSM target parameters are a rated power of 60 kW at 2768 RPM and a theoretical maximum speed of 13,500 RPM. The motor has 48 slots, 8 poles, an outer stator stack diameter of 264 mm, and an axial length of 50.8 mm. Optimization goals are the torque maximization and torque ripple minimization with the minimum efficiency constraint. The optimization procedure includes Latin hypercube sampling, sensitivity analysis, the Kriging method for the creation of a surrogate model and the following GA. Three rotor pole dimensions are varied, which resulted in a higher torque by 8.2% and a lower torque ripple by 28.6%.

A 48-slot 8-pole WRSM with a harmonic exciter is optimized and prototyped in [85]. The rated power is 746 W at 900 RPM, the maximum speed is 3750 RPM, the outer stator stack diameter is 177 mm, and the axial length is 80 mm. Similarly to [83,84], a combination of Latin Hypercube sampling, the Kriging method, and a GA is employed. Torque maximization and torque ripple minimization are the objectives. Three rotor pole geometric parameters are adjusted. Harmonic winding is optimized as well. The torque ripple is reduced from 79.6% to 18%; however, the average torque also reduces from 8.44 N∙m to 7.83 N∙m with a minor loss of efficiency.

Article [86] deals with a 48-slot/8-pole WRSM with an outer stack diameter of 269 mm and axial length of 100 mm. The rated power is 64 kW at 1800 RPM, and the rated torque is 340 N∙m. The impacts of non-grain-oriented and grain-oriented magnetic steels on the WRSM’s performance are evaluated. The optimization is based on LHS, followed by a meta-model and GA. The goal is to maximize power density and efficiency, constraining the torque ripple and terminal voltage. Seven stator and rotor geometrical parameters are manipulated.

The mechanical design of a WRSM is provided in detail in [87]. It is emphasized that rotor end windings bend towards the stator, being affected by centrifugal forces at high speeds. Particularly, it causes much higher stress levels at the edges of the rotor core in comparison with the central area of the core. Aging of the winding resins and impregnations can make it even worse and cause damage to the machine. The rotor end plates are made from a glass-filled plastic, the end caps are made from a non-magnetic alloy, and the rotor slot spacers are designed to avoid the aforementioned issues. Also, the shape of the pole in the stress concentration areas is modified, which affects the transitions from the rotor tooth to the yoke and from the pole shoe to the pole tooth. It is confirmed by FEA that 15,000 RPM can be reached for pure electrical excitation and 13,500 RPM for hybrid excitation because the magnets incorporated in the pole shoes weaken the rotor structure.

A typical structure of an optimization procedure based on the reviewed sources is shown in Figure 5.

Table 3. Summary of WRSM optimization research [67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86].

Application	Optimization Algorithm	Motor Ratings	Ref.
Off-highway electric hybrid truck	Nelder–Mead	Rated power 370 kW, maximum torque 8833 N∙m, CPSR 10:1	[67]
Hybrid electric vehicle	Solid Isotropic with Material Penalization	Rated power 92.5 kW, rated torque 221 N∙m	[69]
Electric vehicle	Physics-informed Bayesian	Rated power 80 kW, rated torque 250 kW	[70]
Hybrid electric vehicle	Differential Evolution	Rated power 160 kW, maximum torque 280 N∙m, CPSR 2.17:1	[71]
Electric vehicle	Global Response Search	CPSR 2.33:1	[72]
Battery electric vehicle	Genetic Algorithm	Rated power 75.5 kW, maximum torque 350 N∙m, CPSR 3.22:1	[73]
Electric vehicle	Differential Evolution	Rated power 55 kW, peak torque 190.55 N∙m, CPSR 3:1	[74]
Passenger electric vehicle	Genetic Algorithm	Rated torque 200 N∙m	[75]
Passenger electric vehicle	Evolutionary Algorithm	Rated power 55 kW, maximum power 190 kW	[76]
Passenger electric vehicle	Genetic Algorithm	Rated power 75.4 kW, rated torque 150 N∙m, CPSR 2.58:1	[78]
Electric vehicle	Genetic Algorithm	Rated power 80 kW, rated torque 250 N∙m, CPSR 2.58:1	[79]
Electric vehicle	Genetic Algorithm	Maximum power 326 kW, CPSR 3:1	[80]
Electric vehicle	Genetic Algorithm	Maximum torque 150 N∙m	[81]
Electric vehicle	Evolutionary Algorithm	Maximum torque 50 N∙m	[82]
Electric vehicle	Genetic Algorithm	Rated power 5 kW, rated torque 31.83 N∙m	[83]
Passenger electric vehicle	Genetic Algorithm	Rated power 60 kW, rated torque 207 N∙m, CPSR 4.88:1	[84]
Electric vehicle	Genetic Algorithm	Rated power 746 W, rated torque 7.91 N∙m, CPSR 4.17:1	[85]
Electric vehicle	Genetic Algorithm	Rated power 64 kW, rated torque 340 N∙m	[86]

provides a summary of the WRSM optimization studies discussed in this section. Note that since the literature sources contain different information on the WRSM ratings, Table 3 in some cases only indicates the motor power, while in other cases more complete information is provided.

5. Brushless WRSM

5.1. Brushes and Slip Rings Excitation

Despite the advantages of WRSMs discussed in the previous section, the application of this technology in electric vehicles imposes certain challenges. One of the main challenges is to ensure a reliable power supply to the rotor field winding. The known production traction WRSMs mentioned in Section 1 use a brushed contact, which is placed in a chamber separated by a sealing from the internal volume of the motor in which the electromagnetic core is placed. This prevents conductive dust generated by the brushes from entering the potentially explosive area. However, the use of a brush assembly reduces reliability, introduces the need for periodic maintenance of the brush assembly, and increases the size of the motor [88]. Therefore, a review of brushless excitation techniques is presented below.

5.2. WRSMs Equipped with a Separate Brushless Exciter

Various variants of the brushless exciter circuit for WRSMs are known [89]. The choice of a specific circuit is determined by the requirements of the target application. In traction motors, brushless exciters based on a rotating transformer are usually used, which, unlike exciters based on synchronous and asynchronous machines, provide power to the field winding over the entire range of rotation speeds, even at zero speed [90,91,92]. An example of a circuit diagram of such a brushless exciter is given in [92]. This circuit, shown in Figure 6, consists of a DC-AC converter, which outputs a single-phase pulse-width modulated (PWM) AC voltage, a single-phase rotating transformer, and a rectifier (diode bridge), the output of which is connected to the field winding of the WRSM. Figure 7a,b show the design options for a rotary transformer presented in [91,92].

Inductive contactless power transfer is employed in [78] and the device is placed in the hollow shaft. A full bridge inverter was used on the primary side and a bridge active rectifier on the secondary side. Maximum efficiency of 95% was achieved. A detailed device design procedure is provided in [93]. Rotating transformer design is elaborated in [94], including the required power electronics on the primary and secondary sides. The prototype of the brushless exciter was built and tested. An important point of quick rotor demagnetization was raised in [94], as it can ensure fast elimination of short-circuit torque after a hypothetical failure in the stator winding. Later, it resulted in patent [95], describing a ‘kill switch’ in the rotor winding that can be controlled remotely. Another solution is offered in [95,96], where no remote control is required for enabling the rotor demagnetization mode as the rotor circuitry activates the switching element when the primary side inverter stops working.

Figure 7. Design options for rotational transformers: (a) based on a ferrite core [91,97]; (b) based on a segmented core employing laminated electrical steel [92].

Figure 7. Design options for rotational transformers: (a) based on a ferrite core [91,97]; (b) based on a segmented core employing laminated electrical steel [92].

An inductive brushless exciter is employed in [64]. The frequency of the power transfer is 100 kHz. Full design and prototyping procedures are given in [98]. Similar construction with partially different parameters was used in [99,100]. Another example of inductive contactless excitation is given in [97]. The maximum efficiency reached 89.5%.

A brushless capacitive exciter is elaborated in [74,76,77] with the suggested application for electric mobility (Figure 8).

5.3. Synchronous Machines with Harmonic Excitation

Since both the use of slip rings and a separate brushless exciter to power the field winding of a WRSM have their pros and cons [88], some researchers also propose the use of excitation systems integrated into the electromagnetic core of the main machine. Therefore, the compactness of a brushless WRSM can be increased by combining the main electromagnetic core with a brushless excitation system. Such harmonically excited synchronous machines (HESMs) can be used in industrial generators [101,102].

As with separate brushless exciters, not all harmonic excitation techniques are suitable for traction applications. Harmonic excitation schemes that provide starting torque close to the rated one are preferred [103]. Otherwise, solutions such as the use of reluctance torque must be used for zero-speed starting [104].

In terms of the stator design, HESMs can be divided into machines with a separate excitation winding on the stator and without a separate excitation winding on the stator, as shown in Figure 9. In terms of rotor design, HESMs can be divided into machines with a single-phase and multi-phase harmonic excitation winding on the rotor, as shown in Figure 10.

A separate harmonic exciter winding on the stator is fed through a separate converter (see Figure 9a). The advantage of this solution is the absence of the need to inject additional harmonic components into the phase currents of the armature winding. In addition, in the case of supplying the harmonic stator winding with alternating current, excitation of the machine at zero speed is achievable. On the other hand, the stator volume is used less efficiently, since it is occupied not only by the main armature winding, but also by the harmonic exciter winding. The reduction in specific torque calls into question the advisability of using such machines, rather than, for example, homopolar inductor machines with a field winding on the stator [67]. In addition, an additional two-stage frequency converter is required to supply the harmonic stator winding [107].

Among the built-in harmonic excitation systems without a separate exciter winding on the stator, there are options with an additional semiconductor converter connected to the neutral point of the stator armature winding, which is used to inject zero sequence into the stator phase currents (Figure 9b). The advantages of this approach are the controllability of the injected current component and a simple and reliable connection scheme for the rotor windings [108]. In the general case, a zero-sequence injection requires two full three-phase inverters: one to form the main voltage component, the second to inject the excitation component [108,109] (Figure 9b). This option can also be considered as an independent power supply for each phase of the machine from a single-phase full-bridge inverter. An obvious disadvantage of this solution is the high cost of the power electronic converter.

Design options with a simplified injector circuit are also presented (Figure 9c). Thus, in [109] an injector circuit is proposed in which only two controllable switches are connected to the neutral point (half-bridge single-phase inverter). In [110], a connection to the neutral point without controlled switches with only two diodes is proposed.

Most of the known HESM designs use a single-phase harmonic winding on the rotor [89,111] (Figure 10a). In [109], an HESM design with a single-phase harmonic winding and zero-sequence current injection in the form of the third spatial harmonic is presented. An experimental comparison of the HESM’s performance using the proposed harmonic excitation and conventional DC excitation is presented.

Study [112] shows the application of zero-sequence injection excitation for an aircraft starter—generator to reduce its dimensions. The application of a dual three-phase stator winding is shown to ensure the machine starts up in the motor mode at any rotor position. An experimental test is presented, including the start mode and generator operation under different loads. In [113], an excitation system based on third-harmonic current injection using two inverters with 180° phase shift is considered. The effectiveness of the approach is verified using FEA. In [114], various HESM configurations are compared when operating with third-harmonic current injection. It is shown that the machine configuration can be optimized for operation with a particular injection scheme.

In [115], an analysis of an HESM with a single-phase harmonic winding excited by two inverters connected to both terminals of the stator winding, controlled in such a way that the input armature current generates an additional third harmonic current component, is presented. Verification of the proposed approach based on FEA is provided. In [116], a finite element analysis of an HESM with an excitation system based on a dual-frequency inverter is presented. A significant reduction in torque ripple (up to 14%) is shown, compared to previously developed HESMs.

In [117], an HESM design with a brushless harmonic exciter and a hysteresis current controller is considered. It is shown that an increase in the frequency of the injected zero-sequence harmonic current (magnetomotive force) induces EMF in the harmonic winding, which allows for an increase in the average torque and efficiency of the machine. However, this solution has the following disadvantage: low inductive coupling of the single-phase harmonic winding with the stator winding at certain rotor positions.

The main disadvantage of the design presented in [117] is the inefficient energy transfer to the harmonic exciter winding at all rotor positions, which reduces the electromotive force (EMF) and makes it difficult to start the motor. The slot sides of the harmonic winding, placed in the slots of the field winding, make a small contribution to the output EMF, and for efficient energy transfer, the angular size of the rotor sub-tooth should be approximately equal to the pole pitch of the spatial third harmonic. At the same time, the entire rotor tooth pitch should be approximately equal to the period of the spatial third harmonic, approximately 120 electrical degrees. This limits the optimization possibilities of this design. A similar HESM design with zero-sequence excitation is presented in [118]. The HESM design with DC excitation of the stator is presented in [110,119].

An HESM with a multi-phase concentrated stator winding and a high-frequency current injection excitation system with a multi-phase harmonic exciter winding on the rotor for use in automotive applications is presented in [103] (see Figure 10b). It is shown that high-frequency current injection using a multi-phase harmonic winding allows for an avoidance of significant torque ripple. At the same time, the disadvantage of the design is the high cost of the inverter since each phase is fed by a separate single-phase full-bridge inverter.

In [105], an HESM design is presented that uses a stator with five or more phases to generate two independently controlled rotating magnetic fields at different spatial harmonics, which provides independent magnetic coupling between the phases of a multiphase harmonic exciter winding (Figure 10c).

In [106], a design of an HESM with a two-phase harmonic winding is proposed, in which the stator winding is inductively coupled to the phases of the harmonic winding at all rotor positions, which increases the excitation efficiency (see Figure 10d). In addition, an improved HESM design with a two-phase harmonic exciter winding is presented, which provides good inductive coupling of the stator winding with the harmonic winding phases in all rotor positions, therefore increasing the excitation efficiency and allowing the machine to start at any rotor position. In the presented design, only one controlled lower switch is used in the injector circuit, compared, for example, with [117], where two controlled switches are used in the injector circuit. In addition, in [106], an HESM model is presented considering PWM.

Table 4. Various WRSM brushless excitation methods suitable for traction applications.

Brushless Excitation Method	Ref.
Separate brushless exciter
Rotary transformer	[90,91,92,93,94,95,96,97]
Capacitive coupler	[74,76,77]
Built-in brushless excitation
Injected zero-sequence harmonic MMF excitation	[104,106,108,109,112,113,114,115,117]
Winding spatial harmonic MMF excitation	[103,105,106,107,108,109,110,111,112,113,114,115,116,117,118]

summarizes the brushless excitation methods proposed in the literature for traction WRSMs discussed in this section.

Table 5. Traction applications of brushless WRSMs [90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106].

Application	Brushless Excitation Method	Motor Ratings	Experimental Prototype	Ref.
Separate brushless exciter
Hybrid electric vehicle	Rotary transformer	Peak torque 310 N∙m, peak power 70 kW, CPSR 3.25:1	Yes	[90]
Hybrid electric vehicle	Rotary transformer	Peak torque 280 N∙m, peak power 160 kW, CPSR 2.4:1	Yes	[94]
Mild hybrid electric vehicle	Rotary transformer	Peak torque 40 N∙m, rated power 20 kW, CPSR 2:1	Yes	[98,99]
Electric mini cargo truck	Rotary transformer	Rated torque 118 N∙m, rated power 22 kW, CPSR 1.55:1	No	[97]
Built in brushless excitation
Electric vehicle	Injected zero-sequence harmonic MMF excitation	1 kW	No	[104,114]
Starter-generator	Injected zero-sequence harmonic MMF excitation	1.5 kW	Yes	[112]
Electric vehicle	Injected zero-sequence harmonic MMF excitation	3 kW	No	[117]
Electric vehicle	Winding spatial harmonic MMF excitation	Peak torque 86 N∙m, peak power 32 kW, CPSR 3.27:1	No	[105]
Electric vehicle	Injected zero-sequence harmonic MMF excitation	2 kW	No	[106]

shows specific applications of brushless traction motors based on the reviewed literature sources. Note that since the literature sources contain different information on the HESM ratings, Table 5 in some cases only indicates the motor power, while in other cases more complete information is provided.

6. Discussion

6.1. Discussion of Design Optimization Methods

Design optimization of traction WRSMs has been covered in many studies with the features briefly summarized in

Table 6. Optimization features.

Feature	Ref.
Coupled electromagnetic and mechanical constraints	[69,71,74,75,79,82]
Coupled electromagnetic and thermal constraints	[69,73,74,75,79]
Optimization for multiple operating points (driving cycle)	[67,71,72,73,75,77,80,82]
Experimental validation presented	[74,75,78,85,87]

. It can be observed that various versions of genetic algorithms dominate in performing such analysis. Mechanical and thermal limitations in multidomain optimization can be imposed before the optimization procedure to define the available design space, or directly in the optimization cycle, or after the optimization to validate the feasibility of the obtained candidates. All these approaches are presented in the literature; however, it can be concluded that the common practice is to reflect thermal limits indirectly via the current densities or verify thermal performance when the optimization is completed.

6.2. Discussion of Brushless Excitation Techniques

Inductive and capacitive exciters feature a compact design, high reliability, and relatively high efficiency, enabling brushless power transfer to the field winding. Various inductance and capacitance arrangements in the primary and secondary circuits improve their performance. Compared to brushed WRSMs, WRSMs with brushless exciters have a higher cost and complexity due to the exciter itself and its associated power electronics.

It can also be concluded that the use of WRSMs with built-in harmonic excitation systems is a promising solution for traction applications, especially those without a separate harmonic winding on the stator and those requiring only a single inverter. This excitation method does not significantly reduce the torque density of the machine and allows for effective excitation levels even at low speeds, down to zero speed. From the point of view of reducing the inverter cost, the use of simplified injector circuits that allow for a reduction in the number of controlled switches is desirable. From the point of view of improving starting capability, the use of a multiphase harmonic winding, which provides good inductive coupling with the stator winding at any rotor position, is recommended.

However, the predominance of theoretical studies in this field and the lack of evidence of brushless excitation use in mass-produced transport vehicles indicate that certain challenges remain unresolved, limiting the application of brushless and, particularly, harmonic excitation in traction drives. These challenges might be addressed in the future through further improvements in design and control techniques.

6.3. Discussion of Comparison of WRSMs and Other Motor Types

PMSMs demonstrate the highest torque density and efficiency among the considered motor types, at least within the constant-torque region. Numerous studies, however, emphasize their drawbacks in applications with the wide constant-power speed range: low efficiency and, in particular, a low power factor, which lead to increased phase current at speeds above the rated speed. Their high cost is caused by the use of rare-earth magnets. On the other hand, PMSMs benefit from relatively simpler rotor cooling systems, as their rotor losses are significantly lower compared to WRSMs and IMs.

IMs have been widely applied in traction for decades and are valued for their simplicity and robustness. This type offers moderate performance in terms of torque density, efficiency, and power factor. Cooling of heavily loaded traction IMs is especially challenging due to the presence of significant rotor electrical losses in the squirrel cage.

SynRMs without assistive permanent magnets are rarely considered for traction applications because of their low torque density, poor power factor, which leads to higher inverter power and cost, and narrow CPSR. Several studies note that adding assistive ferrite magnets can considerably improve performance, though not to the level of PMSMs.

The reported results show that WRSMs and PMaSynRMs achieve comparable specific characteristics and efficiency, yet each of them has its inherent limitations: PMaSynRMs may have a reduced power factor and draw higher current in certain conditions, while WRSMs require rotor excitation to be implemented.

A main challenge of WRSMs is that achieving performance comparable to PMSMs often requires a complex rotor cooling system. Another challenge in this way is the limited mechanical strength of the salient-pole rotor, which heavily restricts the maximum speed.

Overall, the literature reflects a consensus that WRSMs represent a balanced solution, offering relatively low cost, decent traction characteristics, and high efficiency across a wide speed range. Cooling system requirements for WRSMs are lower than those of IMs due to reduced rotor losses, but higher than in PMSMs. The presence of a rotor field winding enables full control of excitation flux but introduces the challenge of supplying power to a winding located on the rotating part of the machine, which is widely highlighted in published works.

7. Conclusions

This article reviews the development and prospects of wound-rotor synchronous motors (WRSMs) for traction applications. Compared to permanent magnet motors, WRSMs do not require the use of expensive rare-earth materials, providing fully controllable flux, a wide constant-power speed range, and comparable performance.

Advances in optimization techniques have improved the performance of WRSMs, although challenges remain in rotor cooling, mechanical strength, and providing a reliable and compact excitation system. Brushless and harmonic excitation schemes show promise in terms of improved reliability, reduced size, and reduced maintenance costs, but their commercial implementation is still limited. Overall, WRSMs represent a balanced and robust solution, particularly for applications requiring high efficiency over a wide range of operating conditions.